Evaluating the Travel Burden of the 2019-2020 NBA Schedule

A Presentation by John Sugden

Full Disclosure

  • I’m a poser
    • no favorite team, no skin in the game
  • I’m a nerd
    • love the stats more than the games themselves
  • I like to travel
    • BUT, it can be exhausting

My Idea

  • Evaulate the travel burden of each NBA team this upcoming season
    • 30 NBA teams
    • Each team plays 82 games per year
    • Each team plays at each stadium at least once

30 Teams, 29 Stadiums

Distance Between Stadiums

  • Identified the nearest international airport to each NBA stadium
  • Looked up the stadium’s geo-coordinates
  • Used R to measure the distance
    • geosphere package
    • distm() function
    • distVincentyEllipsoid parameter (“great circle distance”)

Distance Between Stadiums

Evaluating Existing Schedule

2019-2020 Schedule, Visualized

Approximate Seasonal Air Travel

Utah Jazz

Sacramento Kings

Cleveland Cavaliers

Expected vs Actual: Simple Heuristics

  • Teams near the coast expected to travel more than team in the midwest
  • Rang true for:
    • Portland
    • Sacramento
    • Indiana
    • Chicago
  • Outliers:
    • Utah
    • Los Angeles (Lakers and Clippers)
    • Memphis

Expected vs Actual: Quantitative Approach

  • Normalize average distance from stadiums (“expected”)
  • Normalize total distance traveled (“actual”)
  • Compare the two normalized metrics

Expected vs Actual Burden

We Found a Problem…

  • I’ve pointed out the Uneven travel burden for NBA teams
  • NBA could fix it, if they wanted to, but distance isn’t the NBA’s only concern when setting schedules

Fine, I’ll Do it Myself

Optimizing Schedule Burden

  • Use operations research methods to propose more parity in annual travel burden
  • Relax some assumptions in lieu of practicality

NBA Schedule Constraints

Each team plays:

  • 82 games a year
    • 41 at home and 41 on the road
  • team in their division 4 times per season
  • team in their conference 3-4 times per season
  • team in opposing conference 2 times per season
  • in each stadium at least once per season

Relaxed assumptions

  • Will keep the intra-division games as-is
  • Will keep the opposing conference games as-is

Thus, will only model the 3-4 intra-conference games

Methods

Results Chart

Improvement Chart

Future Considerations

Questions?